| Issue |
ITM Web Conf.
Volume 47, 2022
2022 2nd International Conference on Computer, Communication, Control, Automation and Robotics (CCCAR2022)
|
|
|---|---|---|
| Article Number | 02040 | |
| Number of page(s) | 9 | |
| Section | Algorithm Optimization and Application | |
| DOI | https://doi.org/10.1051/itmconf/20224702040 | |
| Published online | 23 June 2022 | |
Text classification model based on CNN and BiGRU fusion attention mechanism
School of Computer Science and Technology, Harbin University of Science and Technology, Heilongjiang, China
* Corresponding author: hrbustchl@163.com
This model proposes a text classification model with deep learning algorithm, which combines the characteristics of Convolutional Neural Network (CNN) and Gate Recurrent Unit (GRU) in cyclic neural network, extracts local and global features of text feature words respectively, and calculates the importance of words to text classification task after fusing attention mechanism (Attention). Make the model focus on the feature words with high weight. Through the fusion of models, the accuracy of text classification is improved. The experimental results on IMDB film review dataset, Fudan University Chinese dataset and THUCNews dataset show that the proposed model has different degrees of improvement compared with the previously proposed models based on CNN, or LSTM and related fusion models in terms of accuracy, recall rate and F1 value.
Key words: Text categorization / Deep learning / Convolution neural network (CNN) / Gate recurrent unit (GRU) / Attention
© The Authors, published by EDP Sciences, 2022
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.